[dpdk-dev] [PATCH v3] eal/ppc64: improve rte_rdtsc() with __ppc_get_timebase()

[Thomas Monjalon]

Any review please?

25/03/2020 01:13, Thinh Tran:
>   __ppc_get_timebase() reads and returns the current value of the Time 
>   Base Register. It's more efficient as it uses the processor’s time  
>   base facility directly 
> 
>   the DPDK on FreeBSD currently is not supported on Powerpc64, it should
>   be safe to include the sys/platform/ppc.h
> 
> 
> Signed-off-by: Thinh Tran <thinhtr at linux.vnet.ibm.com>
> ---
>  .../common/include/arch/ppc_64/rte_cycles.h   | 28 ++-----------------
>  1 file changed, 2 insertions(+), 26 deletions(-)
> 
> diff --git a/lib/librte_eal/common/include/arch/ppc_64/rte_cycles.h b/lib/librte_eal/common/include/arch/ppc_64/rte_cycles.h
> index 8f2e98642..871f9b6e4 100644
> --- a/lib/librte_eal/common/include/arch/ppc_64/rte_cycles.h
> +++ b/lib/librte_eal/common/include/arch/ppc_64/rte_cycles.h
> @@ -14,6 +14,7 @@ extern "C" {
>  
>  #include <rte_byteorder.h>
>  #include <rte_common.h>
> +#include <sys/platform/ppc.h>
>  
>  /**
>   * Read the time base register.
> @@ -24,32 +25,7 @@ extern "C" {
>  static inline uint64_t
>  rte_rdtsc(void)
>  {
> -	union {
> -		uint64_t tsc_64;
> -		RTE_STD_C11
> -		struct {
> -#if RTE_BYTE_ORDER == RTE_BIG_ENDIAN
> -			uint32_t hi_32;
> -			uint32_t lo_32;
> -#else
> -			uint32_t lo_32;
> -			uint32_t hi_32;
> -#endif
> -		};
> -	} tsc;
> -	uint32_t tmp;
> -
> -	asm volatile(
> -			"0:\n"
> -			"mftbu   %[hi32]\n"
> -			"mftb    %[lo32]\n"
> -			"mftbu   %[tmp]\n"
> -			"cmpw    %[tmp],%[hi32]\n"
> -			"bne     0b\n"
> -			: [hi32] "=r"(tsc.hi_32), [lo32] "=r"(tsc.lo_32),
> -			[tmp] "=r"(tmp)
> -		    );
> -	return tsc.tsc_64;
> +	return __ppc_get_timebase();
>  }
>  
>  static inline uint64_t